Amazon has released new machine learning tools with a wide range of applications for the healthcare space.
The new Amazon Elastic Inference service harnesses a scalable GPU model, enabling customers to choose the amount of computing power they need with the option to scale up or down as demands dictate.
GPUs provide a tremendous surge in machine learning power for massive datasets such medical imaging. This system provides cost-effective computing power to serve predictions to medical researchers at a scalable level.
Meanwhile, new additions to Amazon Sagemaker, a machine learning framework hosted on AWS, have healthcare applications as well.
SageMaker Ground Truth uses active learning and can be “trained” in real time to perform data labelling and processing, such image and text classification, or object detection. Labelling a dataset of millions of documents is a resource intensive task that often stands in the way of allowing machine learning to serve predictions.
SageMaker Neo optimizes machine learning instances during the training phase, allowing health organizations to wring maximum efficiency out of whatever hardware architecture they might be using for their prediction environment.
And SageMaker RL is a reinforced learning program which can be used where “building a prior dataset would either be infeasible or prohibitively expensive,” according to Amazon’s blog post. Reinforced Learning uses continuous feedback to constantly improve problem solving and make “increasingly relevant actions,” officials said.
WHY IT MATTERS
As data proliferates in healthcare organizations, the need to quickly and efficiently process it is vital both to improving patient outcomes as well as developing revenue streams from the information captured. Advances in computing technology, especially GPUs, have allowed machine learning to grow by leaps and bounds.
However, the physical architecture that enables these services can be prohibitively expensive for a healthcare system. Amazon’s cloud-based offerings enable data workers to harness the power of the advanced AI in a scalable and cost-effective environment.
“With petabytes of unstructured data being generated in hospital systems every day, our goal is to take this information and convert it into useful insights that can be efficiently accessed and understood,” Anish Kejariwal, director of software engineering for Roche Diagnostics Information Solutions, said in an Amazon blog post.
THE LARGER TREND
Amazon’s cloud computing and machine learning offerings have been making increased inroads into the healthcare space. It has developed healthcare-specific tools to analyze patient records and a more and more of its portfolio is HIPAA-eligible as it continues to home in on healthcare. Machine learning is being deployed throughout the industry and in some cases is dramatically reducing the amount of time spent by clinicians.
Earlier this week, AWS also unveiled Amazon Comprehend Medical, new HIPAA-eligible machine learning tool, enabling developers to process unstructured medical text and spot specific data such as diagnosis, treatments, dosages, symptoms and more.
So much of today’s healthcare data is unstructured medical text – written notes and audio transcripts, prescriptions, pathology and radiology reports – exists as hard-to-mine unstructured data. With the new Comprehend Medical tool, “developers only need to provide unstructured medical text,” said Amazon in a blog post. “The service will ‘read’ the text and then identify and return the medical information contained within it.
Moreover, the technology will also highlight protected health information, officials said, and no data processed by the service is stored. (Comprehend Medical is covered under AWS’ business associate agreement.)
With no machine learning experience required – there are no models to train – and the ability to be integrated with existing services via API, the tool could be a valuable resource for clinical decision support, revenue cycle management, pop health, clinical trials and more, Amazon said.
ON THE RECORD
“The process of developing clinical trials and connecting them with the right patients requires research teams to sift through and label mountains of unstructured medical record data,” Matthew Trunnell, CIO at Fred Hutchinson Cancer Research Center, said in the blog post.
“Amazon Comprehend Medical will reduce this time burden from hours per record to seconds,” he said. “This is a vital step toward getting researchers rapid access to the information they need when they need it so they can find actionable insights to advance lifesaving therapies for patients.”
Benjamin Harris is a Maine-based freelance writer and and former new media producer for HIMSS Media.
Focus on Artificial Intelligence
In November, we take a deep dive into AI and machine learning.